基于参数分析的物流网络优化问题研究

详细信息    查看全文 | 推荐本文 |

英文篇名：Research on Logistics Network Optimization Based on Parameter Tuning 作者：夏振喜 英文作者：Xia Zhenxi;School of Logistics Engineering,Wuhan University of Technology; 关键词：参数分析 ; 带边中断动态网络最大流问题 ; irace ; 贪婪随机自适应搜索算法 英文关键词：parameter tuning;;maximum total flow with flexible arc outages;;irace;;greedy stochastic adaptive search algorithm 中文刊名：WLJS 英文刊名：Logistics Technology 机构：武汉理工大学物流工程学院; 出版日期：2019-02-25 出版单位：物流技术 年：2019 期：v.38;No.389 基金：国家自然科学基金资助项目(71501152) 语种：中文; 页：WLJS201902010 页数：8 CN：02 ISSN：42-1307/TB 分类号：46-52+90

摘要

介绍了带边中断动态网络最大流问题,在运用含参数的贪婪随机自适应搜索算法对该问题进行求解时,其算法内部参数设置问题未能得到较好的解决,针对这一问题,采用目前比较热门的参数分析工具irace对求解该问题算法的参数进行分析,寻求算法的最佳参数设置,通过实验设计与分析表明,irace在参数分析的基础上得出的结果要比在默认参数下求得的结果更好,可以根据irace分析出的结果及时更改原来GRASP算法中的默认参数,使得最终结果更接近最优解。
        This paper introduces the problem of maximum total flow with flexible arc outages.When an attempt is made to solve this problem by way of the greedy stochastic adaptive search algorithm that includes parameters,the setting of the internal parameters would interfere with the solution process.To deal with this issue,the most popular parameter analytic tool at present,irace,is used to analyze the parameters of the algorithm to seek the optimal setting.By designing and performing an experimental study,it finds out that the results obtained using irace on the basis of parameter tuning are better than those obtained with the default parameters.Moreover,the default parameters in the original GRASP algorithm can be changed in a timely manner against the analytic result of irace,so that the final result is closer to the optimal solution.

引文

[1]Adenso-Diaz B, Laguna M. Fine-Tuning of Algorithms Using Fractional Experimental Designs and Local Search[J].Operations Research,2006,54(1):99-114.
    [2]López-Ibánez M,Dubois-Lacoste J,Cáceres L P,et al.The irace Package:Iterated Racing for Automatic Algorithm Configuration[J].Operations Research Perspectives,2016,3:43-58.
    [3]Hutter F,Hoos H H,Leyton-Brown K,et al.ParamILS:An Automatic Algorithm Configuration Framework[J].Journal of Artificial Intelligence Research,2014,36(1):267-306.
    [4]Hutter F, Hoos H H, Leytonbrown K. Sequential Model-Based Optimization for General Algorithm Configuration[A]. Learning and Intelligent Optimization-,International Conference[C].2011.
    [5]陈一昭,姜麟.蚁群算法参数分析[J].科学技术与工程,2011,11(36):80-84.
    [6]刘伟.蚁群算法参数分析与组合优化设置研究[J].电脑与信息技术,2011,19(1):10-12.
    [7]Grefenstette J J.Optimization of control parameters for genetic algorithms[J]. IEEE Transactions on systems, man, and cybernetics,1986,16(1):122-128.
    [8]Hutter F,Hoos H H,Leyton-Brown K,et al.ParamILS:an automatic algorithm configuration framework[J]. Journal of Artificial Intelligence Research,2009,36:267-306.
    [9]Boland N,Kalinowski T,Waterer H,et al.Scheduling arc maintenance jobs in a network to maximize total flow over time[J].Discrete Applied Mathematics,2014,163(1):34-52.
    [10]Pearce R H,Forbes M.Disaggregated benders decomposition for solving a network maintenance scheduling problem[J].Journal of the Operational Research Society,2018:1-14.
    [11]刘茜,郑澜波.基于Benders分解的煤炭供应链设备维护计划决策[J].物流技术,2017,(12):128-132.
    [12]Xu Q,Liu X,Zheng L.A Hybrid Tabu Search and LP Heuristic for the Maximum Total Flow with Flexible Arc Outages[J].Matheuristics,2016:13.